Question 1

How has educational achievement in math and science changed over time?

In the earlier version of this figure, I plotted all of the available data, and showed math and science achievement separately. This was helpful to get a feel for the data, but it was difficult to get the main takeaways from the figure with a quick glance. I tried a few different ways of making the plot, and peer review helped me to decide which one was the best way of showing the data. For the final version, I made a couple of major changes. First, I dropped the lines that showed the proportion of students below basic and below proficient. Since the plot already showed the proportion of students above basic and above proficient, this was redundant information. This also helped to reduce the number of colors happening in the plot, as it no longer needs a four-color legend. I also switched to a color palette that had more distinct colors, and checked the new plot with cvd_grid() to see if the new colors were colorblind-friendly. Additionally, I put both math and science achievement on the same plot, making them easier to compare. For the final plot, I used both color and line types to convey the information in the plot. Color is used to differentiate between subjects, while line types (solid versus dashed) show the different achievement levels.

Question 2

How does educational achievement in math and science differ across states?

For this set of plots, I originally used the geom_sf() function to make a more traditional map of the US. However, geom_sf() took an incredibly long time to load, and resulted in some messages/documentation of the package loading that still showed up in the knitted document. Additionally, the traditional map makes it difficult to see what’s happening in geographically smaller states. To solve those problems, I switched to using the statebins package instead. The statebins package represents each state as a square of the same size, making it easier to see smaller states. The statebins package also labels each state, which is helpful for interpretation. The one issue I ran into with the package was when trying to facet the plots. Originally, I was trying to make a plot for each subject at one timepoint, and facet by achievement level. However, these plots always looked squished, and the state labels were difficult to read. As a solution, I switched to using a dashboard, where I could just make tabs for the different achievement levels. Based on feedback, I also updated the labels on the plots, including the title, subtitle, and legend labels.

Question 3

a. How does funding differ between states?

b. How has funding for education changed over time?

c. How have changes in funding differed between states?

These plots were inspired by the problem of trying to find a feasible way to represent data from all 50 states in the same figure. To make a readable figure with all 50 states represented, I tried to make a circular bar plot. For my first version, I made a circular bar plot based on per-student expenditures averaged over the time period for which I had data. In the first version, the states were just organized alphabetically. Based on feedback, I tried grouping the states by funding quantiles, colored the plot by groups, and added space between each quantile. I also added labels for each quantile, and changed the scaling of the data to make bars fit into the circular plot more smoothly, and adjusted the label alignment to make them easier to read. I also checked whether the default colors for the groupings were colorblind-friendly. While they worked for three of the four types of colorblindness included in the cvd_grid() function, for the fourth version, all groups looked the same. I changed the colors to a colorblind-friendly pallette, and checked the new colors against cvd_grid(). Based on additional feedback, I also sorted the bars within each quantile by their value. For both the circular bar plot and the line plot underneath, I added more informative titles.